| Position | A senior Harvard figure. The release does not yield his title cleanly: on the two sheets that would give it, the word is covered by a bar1 |
| The act | He solicited Epstein for a $2 million naming gift for the Harvard Hillel building, on his own account as quoted in a submission made for Epstein2 |
| Where that account sits | Inside a mitigation package hand-delivered to the State Attorney prosecuting Epstein, during the non-prosecution negotiation3 |
| Correspondence | A sustained personal correspondence with Epstein from August 2009 to at least June 2015, mostly about arranging to meet in Cambridge4 |
| In his own words | ”please ask Jeffrey to come to our apartment—he has been here many times—“5 |
| Died | November 2022, aged 95, so he was dead for nearly three years before the 2025 interview below and could not answer it6 |
| 2025 | Named once in an unsworn DOJ proffer interview of Ghislaine Maxwell, in a passage the transcript leaves unresolved nine lines later. See the section below, which prints the exchange whole7 |
Henry Rosovsky is a Harvard economist and administrator who appears in the released files in two quite different characters. In one he is a fundraiser: by his own account, quoted in a document prepared for Epstein’s defence, he approached Epstein and suggested that a $2 million gift would buy his name on a Harvard building. In the other he is a correspondent, exchanging short notes with Epstein over six years about lunches, coffees and a standing invitation to his apartment in Cambridge.
A third appearance is a single passage in a 2025 interview that this page prints in full and refuses to summarise, because the speaker may have withdrawn it nine lines later and the record never resolves which man she meant.
The Harvard Hillel solicitation
The account is Rosovsky’s, but the page it sits on is not his. A submission prepared on Epstein’s behalf sets the scene in its own voice: that Epstein “has been a benefactor to a number of educational institutions, but perhaps none more so than Harvard University”; that the Jewish community at Harvard needed a new Hillel Center; that “the fundraising campaign was having difficulties when Henry Rosovsky, then ▮ of the Faculty of Arts and Sciences, decided to approach Jeffrey.” The bar is the document’s; it covers his title.2
Then it quotes him. The quotation begins with “At this point”, and the antecedent is the submission’s sentence above, not anything of Rosovsky’s, so the two are given together:2
At this point I approached Jeffrey and suggested that a $2 million gift would allow him to have his name on the building. It seemed to me a tempting proposition. To get your name on a Harvard building is a rarity and the cause was worthy. I do not remember all the subsequent details, but in essence this is what happened.
Jeffrey told me that he would raise the naming gift and that he certainly did not want his name on the building. Leslie Wexner would also participate and he did not want his name on the building. Indeed, he had decided – by himself or jointly with Wexner – that my name should be on the building because I had been associated with the institution for such a long time. I fully recognized the generosity and modesty of this suggestion and greeted it with mixed feelings. I did not feel deserving of this honor but without the Epstein-Wexner participation there would have been no building. In truth, without Jeffrey’s intervention there would have been no new building and that would have been an enormous loss to the community.2
His quoted words stop there. The outcome is stated by the submission’s author, not by him: that the building standing on the Harvard campus is “the Riesman Center for Harvard-Radcliffe Hillel at Rosovsky ▮”, and that the episode “suggests a noteworthy aspect of Jeffrey’s giving: to the extent that it is possible, Jeffrey wants his gifts to be anonymous.”2
Four things the sheet does not say, and this page will not supply. It does not say that $2 million was ever paid: that is the figure Rosovsky says he suggested. It does not say who paid what, only that Epstein “would raise the naming gift” and Wexner “would also participate”. It gives no date for the solicitation, the gift, the construction or the naming. And it does not give his title, which is barred.2
Where this account was being used
The document matters as much as the quotation. It is not a memoir and not a Harvard record: it is pages 75 to 89 of an eighty-nine-sheet Palm Beach State Attorney public-records production, and the submission was delivered to the prosecutor.3
Its transmittal, three sheets earlier, is a hand-written note on the letterhead of Epstein’s Florida counsel:
Barry / Here is Acosta’s number… because Jay Lefkowitz has to follow up w/a call to him to finalize. / Enclosing some fun reading on your Defendant / Jack3
“Barry” is State Attorney Barry Krischer; “Jack” is Jack Goldberger; “Jay Lefkowitz” is Jay Lefkowitz of Kirkland & Ellis, and “Acosta” is the United States Attorney. The note sits immediately after a September 2007 message on the non-prosecution agreement.3
So Rosovsky’s warm account of Epstein’s philanthropy reached the file as mitigation material, handed to the state prosecutor while the federal agreement was being negotiated. Nothing in the release suggests Rosovsky knew that is what it would be used for, and this page does not suggest it either.
The correspondence, 2009 to 2015
The earliest document located is a relay of 26 August 2009 from Epstein’s assistant Lesley Groff, quoting him in the first person: “I am available between 1-3:00 to speak today. If you could pin point a time I will have my partner on the call as well.”4
From 2012 he writes directly, and the notes are short and social. On 16 April 2012, replying to Epstein: “Yes. Next time a bit more one on one but your assembled group was extremely interesting.”4 A month earlier he had cancelled a meeting: “I committed a great boner! I completely forgot that April 15 is my wife’s birthday!!”4 In April 2013 an intended visit fell through and he wrote: “I guess that he couldn’t make appointments with the important people he wanted to see. Ah well…”4 A June 2014 cancellation opens by noting he had seen it coming, “What I thought might happen is happening: the family is gathering at the Cape this Saturday and my presence is required”, and asks that his “profound apologies” be passed to Epstein.4
The clearest statement of the relationship is from 28 May 2015, when Groff asked whether he would be around for a Harvard visit:
Yes, I will be back and have a lunch appointment. But please ask Jeffrey to come to our apartment—he has been here many times— any time after 2 PM. My experience in meeting him at Martin’s institute means sitting there with lost of people and no opportunity to have a real conversation. That does not work for me.5
The dashes are as printed, one long and then two hyphens; “lost of people” is his typing. “He has been here many times” is the page’s clearest evidence of how often the two met, and it is his own sentence. “Martin’s institute” is the Harvard programme run by Martin Nowak, whom Rosovsky names elsewhere in the correspondence, though spelled “Novak”.5 That identification is corroborated from outside the correspondence: asked in 2018 about the sessions at Nowak’s Program for Evolutionary Dynamics, Rosovsky told The Harvard Crimson that he “attended a couple of sessions and lost interest”.8
His name also sits on a call list of 31 May 2011 drawn up for Epstein alongside Dick Merkin, Woody Allen, Marty Seligman, David Gergen and Michael Wolff, and an assistant reported back that she had left a message for him at home. Every contact value on that sheet is barred.9
2025: one passage, printed whole
In July 2025 Ghislaine Maxwell was interviewed over two days by the Deputy Attorney General. The interview was not under oath: the document describes itself as a recorded proffer, and the words oath, sworn and swear appear nowhere in its 263 sheets.7
On the first day she was asked about prominent people in Epstein’s life in the 1990s, and read names from her own notes, listing him as “Henry Rosovsky, who was the provost of Harvard”.7 Some pages later the interviewer returned to that list. What follows is the exchange entire, over two transcript pages, because no shorter extract is honest:
TODD BLANCHE: … And then the — the kind of what — what I called famous friends, but the — the prominent individuals that were in his life in the ’90s. Did — did — does any stick out in your mind as having received massages? All of them.
GHISLAINE MAXWELL: Henry Rosovsky received a massage.
TODD BLANCHE: And why do you — why does that stick out in your memory?
GHISLAINE MAXWELL: Because I saw him in a bathrobe at 71st Street, and he had received a massage, he told me.
TODD BLANCHE: And do you know whether that — whether there was any — whether the masseuse was naked during that massage?
GHISLAINE MAXWELL: I wouldn’t have any idea.
TODD BLANCHE: Do you know whether he —
GHISLAINE MAXWELL: I doubt it. He was like in his 80s.
TODD BLANCHE: Say it again.
GHISLAINE MAXWELL: I doubt it. He was like in his 80s.
TODD BLANCHE: Okay. So — but do you know — notwithstanding his age —
GHISLAINE MAXWELL: Minsky, sorry.
TODD BLANCHE: Say that again.
GHISLAINE MAXWELL: Minsky was another person.
TODD BLANCHE: Do you know whether, for example, President Clinton ever received a massage?7
⚠ “All of them.” is the interviewer’s, not hers. It sits inside his own untagged block, on the same line as his question, and the reporter tags every change of speaker throughout. It is reproduced here because it is part of how the question was put, and it is attributed to him because that is what the transcript shows.7
What this page will not say
It will not say which man she meant. “Minsky, sorry.” interrupts the interviewer mid-question, and “sorry” is the ordinary token of self-correction; against that, “another person” is the ordinary English of addition, the interviewer had just said “we might add some names later”, and she was reading from notes. The transcript never resolves it. The interviewer does not restate either name, does not ask which she means, and moves to President Clinton in the next line. Neither name appears again in the day’s 213 pages, established from the transcript’s own court-reporter word index, which lists “Rosovsky” at two lines and “Minsky” at two lines, all four inside the passage above.7
⚠ And it will not let the reader forget who could not answer. Rosovsky died in November 2022, aged 95. The interview is of July 2025. The passage is therefore an unresolved assertion, made in an unsworn proffer, about a man who had been dead for nearly three years and who has never been asked about it.6
It will not say the Minsky is Marvin Minsky. She names “Marvin Minsky” only on the following day, in a different list, in answer to a different question with no massage in it, and Rosovsky is not mentioned that day at all. Joining the two is an inference across two documents, and the page declines to make it.10
And it will not present the first line as a standing allegation. An assertion the speaker may have withdrawn nine lines later, in an unsworn interview, with no corroboration anywhere in the release, is not a credible allegation against a named man. It is an unresolved passage, and that is all this page claims it to be.
Harvard
The wider record of Epstein’s dealings with the university, covering the Program for Evolutionary Dynamics, the gifts, the 2020 institutional review, and the Hillel building whose naming is described above, is set out at Epstein and Harvard University, which identifies the Hillel building as Rosovsky Hall and carries the 2018 Crimson reporting.8
His name in the record
Measured against the full-text index on 18 September 2026, count(distinct efta_number) over
text_content: “Rosovsky” returns 529 documents and “Henry Rosovsky” 354. He is a sender in
37.11
Two cautions for anyone searching. The reversed form “Rosovsky, Henry” is not a separate query from
“Rosovsky Henry”, because the index splits on commas; the same is true of “From: Rosovsky” and
“From Rosovsky”. And the plain surname is not the ceiling: a prefix search on Rosov* returns 656,
so roughly 127 documents carry a form the exact spellings miss.11 No second bearer of the surname
was found.
What is not established
- His title. Barred on both sheets that would give it. The page declines to supply it from outside the release.1
- Whether the $2 million was paid, by whom, or when. The figure is what he says he suggested.2
- Any date for the solicitation or the naming.2
- Whether he knew what the submission quoting him was for.
- Which man Maxwell meant, and whether the Minsky she named is the person of that name already in this record.710
- Whether he and Epstein met other than in Cambridge. The correspondence is about Cambridge; nothing read here places him elsewhere.
Footnotes
Footnotes
-
The sheets that would give his title bar it. On the character submission the phrase reads “then ▮ of the Faculty of Arts and Sciences”; the other sheet located carries a bar at the same place. In the 2025 interview Maxwell describes him as “the provost of Harvard”, but that is a speaker’s description in an unsworn interview and is not adopted here as his title. ⚠ One of these bars under-covers at 600 dpi; no crop of that line is published and nothing is recovered from it. ↩ ↩2
-
Character submission for Jeffrey Epstein, at pages 75 to 89 of an eighty-nine-sheet Palm Beach State Attorney public-records production, Request No. 19-372. EFTA02857384 p.82 (the submission’s own internal page 8). Read at the image at 400 dpi whole-page and 600 dpi on the quoted lines; the quotation is verbatim at both ends. The dashes in ”– by himself or jointly with Wexner –” are spaced and symmetric on the sheet, where the text layer normalises them. The bar in “then ▮ of the Faculty of Arts and Sciences” and the bar in “at Rosovsky ▮” are the document’s; nothing is recovered from either here. Rosovsky is quoted, not the author: the submission introduces the passage in its own roman voice with “Rosovsky wrote:” and sets his words as an indented italic block, referring to him in the third person on the same page. The underlying writing is not in the release. The solicitation appears once in the release, under this identifier and its Justice Department twin. ⚠ Cite the page, never the run — see 3. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8
-
Transmittal at pages 72 to 74 of the same production, a hand-written note on the letterhead of Epstein’s Florida counsel, undated, sitting immediately after a September 2007 message concerning the non-prosecution agreement. EFTA02857384 pp.72-74. ⚠⚠ This identifier is a MIXED PRODUCTION and must never be cited at run level or described by the character of any one enclosure. Its page 1 is an e-mail of 1 April 2010, and the eighty-nine sheets include documents of wholly different character from the submission, among them material this desk has identified and deliberately not read. The exposure is recorded in full at
_desk/DECISIONS_NEEDED.mdunder decision AB and awaits an editor’s ruling; no page cites anything from those parts of the run. ↩ ↩2 ↩3 ↩4 ↩5 -
Correspondence between Rosovsky and Epstein and Epstein’s staff. The 26 August 2009 relay by Lesley Groff, quoting him in the first person, is EFTA02435663 p.1; his reply of 16 April 2012 is EFTA01873635 p.1; the March 2012 cancellation is EFTA02173238 p.1; the April 2013 note is EFTA02140482 p.1; the June 2014 cancellation is EFTA02102961 p.1. On every one of these sheets the
From:value is in the clear and the sender’s address field is absent from the sheet, not barred. Timestamps printed on these sheets are UTC and are not converted here. The June 2014 note is quoted from its opening rather than from mid-sentence, because the opening is what shows he had anticipated the cancellation. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 -
Rosovsky to Epstein’s office, 28 May 2015. EFTA02078025 p.1. Read at the image. The punctuation is as printed: one long dash before “he has been here many times” and two hyphens after it, and “lost of people” is the sheet’s spelling. The message closes “Cheers, / Henry”; the quotation above stops before the closing. ⚠ This is one of nine copies of the message in the release and the only one that is a sheet Rosovsky sent; the other eight enclose it. A further note of 3 June 2015 gives the apartment’s general location relative to a Cambridge hotel and is not reproduced. The identification of “Martin’s institute” rests on his naming Martin Nowak elsewhere in the correspondence, spelled “Novak”, and is stated on this page as probable rather than established. ↩ ↩2 ↩3
-
Henry Rosovsky died in November 2022, aged 95. Recorded on this wiki at Epstein and Harvard University, which cites a New York Times obituary of 16 November 2022 and states that he “died in 2022 and cannot answer the document”. This page prints an assertion about him made in July 2025, and says so beside it. ↩ ↩2
-
Interview of Ghislaine Maxwell by the Deputy Attorney General, 24 July 2025, transcript page 100 for the antecedent and pages 102 to 103 for the exchange. EFTA02846680 pp.100, 102-103, a copy set at one transcript page per sheet, which is why the viewer page and the transcript page agree. Read at the image. The interview was not under oath: the words oath, sworn and swear appear on no page of the 263-sheet document, which describes itself at p.2 as a recorded proffer. “All of them.” at 102:16 is set flush inside the interviewer’s own untagged block, on the same line as his question, and the reporter tags every change of speaker; it is therefore his and is attributed to him. That neither name recurs is established from the transcript’s own court-reporter word index, which lists “Rosovsky” at 100:19 and 102:17 and “Minsky” at 103:12 and 103:14, and from nothing else. A second copy of the transcript exists at four transcript pages per sheet and is not cited here because the pagination is harder to check. The passage is reproduced nowhere else in the release. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7
-
The Harvard Crimson, 2018, reporting his answer when asked about the sessions at the Program for Evolutionary Dynamics: that he “attended a couple of sessions and lost interest”. Quoted with its citation at Epstein and Harvard University, which also identifies the Harvard Hillel building as Rosovsky Hall and sets out the wider record of Epstein’s dealings with the university. This page takes the quotation from that page rather than from a document it has read itself. ↩ ↩2
-
Call list prepared for Epstein, 31 May 2011, naming Dick Merkin, Woody Allen, Marty Seligman, Henry Rosovsky, David Gergen and Michael Wolff. EFTA01866253 p.1. Read at the image: every contact value on the sheet is barred and none prints; the labels “office”, “cell” and “h)” print without values. The name “Gergen” was confirmed at the image, the text layer’s “Gerson” being an extraction error. The assistant’s report of having left a message for him at home is EFTA01867101 p.1. ↩
-
Interview of Ghislaine Maxwell, 25 July 2025, the second day, transcript page 276, where she names “Marvin Minsky” within a group of scientists associated with Harvard. EFTA02847046. The question that day contains no reference to massages and Rosovsky is not named on that day. Connecting this mention to “Minsky, sorry.” on the previous day would be an inference across two documents and this page does not make it. ↩ ↩2
-
Measurements taken 18 September 2026 against
pages_ftsovertext_content, countingdistinct efta_number: “Rosovsky” 529; “Henry Rosovsky” 354; “From: Rosovsky” 37. ⚠ Because the index splits on commas, “Rosovsky, Henry” is not a distinct query from “Rosovsky Henry”, and “From: Rosovsky” is not distinct from “From Rosovsky”; counts reported for those forms elsewhere are duplicates rather than additions. A prefix search onRosov*returns 656, so about 127 documents carry a form the exact spellings do not reach; those have not been classified. Two candidate misspellings returned zero and therefore close nothing. No second bearer of the surname was found in the documents read for this page. ↩ ↩2